From a Monthly Egress Total to a Sustained Gigabit Figure
Delivery invoices and capacity contracts almost never speak the same language. The traffic report from a CDN or an object store counts what left the edge in gigabytes, while the port, the transit commit and the peering session are all sold in gigabits per second. Before you can tell whether a month of egress justifies a bigger commit — or whether you are paying for capacity you never touch — the volume has to be restated in bits.
That is the whole job of this page: take the GB number off the traffic report, turn it into gigabits, and then decide for yourself what rate it implies over the billing window.
Gb = GB × 8. Worked example: 10,000 GB of egress in a 30-day month is 80,000 Gb. Spread evenly over the 2,592,000 seconds in that month, it works out at roughly 0.0309 Gbps of sustained throughput.What Each Side of the Bill Is Really Measuring
Volume pricing (per GB)
Rate pricing (95th percentile)
Commits and burst headroom
Working a Traffic Report Through the Converter
Enter the billed egress volume
Type the gigabyte total from the usage report into the left field. It is already set to GB, and the box accepts a comma or a dot as the decimal mark and ignores any spaces pasted in with the figure.
Read the gigabit equivalent
The right-hand field updates as you type. Large totals get a space between thousands so a seven-figure gigabit number stays readable, and anything from ten billion upward switches to scientific notation instead of running off the screen.
Copy the bare number into your capacity model
The copy button on each field puts the plain figure on the clipboard — no unit, no spacing — so it drops straight into a spreadsheet cell where you divide by the seconds in the billing period. Pressing Ctrl+C inside a field does the same thing.
Go the other way when the vendor quotes bits
Both fields are editable, so you can type a gigabit figure from a quote on the right and read the byte volume on the left. The swap button flips the pair outright, and the searchable dropdown on either side lets you switch to Gbps, GB/s, Tb or TB when a proposal mixes rate units with volume units.
Monthly Egress Volume and the Bandwidth It Implies
The table walks a month of egress through both steps: multiply by 8 for gigabits, then divide by 2,592,000 seconds for the average rate across a 30-day month. The last column is the reality check — real delivery traffic is never flat, and a peak-to-average ratio near 5:1 is a common planning assumption for consumer-facing content with a nightly prime time.
| Monthly egress | Gigabits transferred | Average rate (30-day month) | Port implied at 5:1 peak-to-average |
|---|---|---|---|
| 1,000 GB (1 TB) | 8,000 Gb | 0.0031 Gbps | ~0.015 Gbps |
| 10,000 GB (10 TB) | 80,000 Gb | 0.0309 Gbps | ~0.154 Gbps |
| 50,000 GB (50 TB) | 400,000 Gb | 0.1543 Gbps | ~0.772 Gbps |
| 100,000 GB (100 TB) | 800,000 Gb | 0.3086 Gbps | ~1.54 Gbps |
| 500,000 GB (500 TB) | 4,000,000 Gb | 1.5432 Gbps | ~7.72 Gbps |
| 1,000,000 GB (1 PB) | 8,000,000 Gb | 3.0864 Gbps | ~15.4 Gbps |
Read from the bottom up, the table explains something that surprises most people the first time they model it: a petabyte a month — a serious content bill — averages barely 3 Gbps. What forces a 10 Gbps port is rarely the volume. It is the shape of the demand curve.
Rate units sit in the same dropdown
Both unit lists are searchable and carry Gbps, MBps and the rest alongside plain GB and Gb, so a per-second quote and a per-month total can be normalised on one page.
Clipboard output your spreadsheet accepts
Copying a result yields the raw number with no unit and no separators attached, so a capacity model never has to strip formatting before dividing by the billing period.
Readable at petabyte scale
Thousands are separated by a space and very large results flip to exponent form, which keeps a year of aggregated egress legible instead of an unreadable run of digits.
Forecast figures stay on your machine
Traffic numbers from an unannounced launch plan or a confidential quote are converted locally after the page loads — nothing you type is transmitted.
Bandwidth Billing Questions
Why does my per-GB CDN invoice tell a different story than my 95th-percentile transit invoice?
They meter two different things. Per-GB billing adds up every byte delivered over the month, so the total tracks volume no matter how it was distributed in time. A 95th-percentile port bills the sustained rate instead: five-minute samples are collected all month, the busiest 5% — roughly 36 hours' worth — are discarded, and the highest remaining sample sets the charge. A spiky launch can double the per-GB bill while barely moving the percentile, and a flat always-on stream does the opposite.
How much headroom should I add on top of the average Gbps this conversion produces?
The average is a floor, never a plan. Measure your own peak-to-average ratio from five-minute interface data across a full month, because it varies enormously by audience: a single-region consumer service with a sharp evening peak often lands near 5:1, while globally spread API or software-update traffic can sit closer to 2:1. Apply your measured ratio to the average, then leave room on top for the failover case where one region's traffic lands on another region's ports.
Does a 10 Gbps commit entitle me to a fixed number of gigabytes each month?
Not the way a data cap does. A rate commit sets what you pay for capacity, not an allowance of bytes, and the theoretical ceiling of a fully saturated link is enormous: 10 Gbps × 2,592,000 s = 25,920,000 Gb, which is 3,240,000 GB in a 30-day month. No real service runs a port flat out, so what a commit actually buys is the right to burst without overage. Convert before you compare offers — a volume deal and a rate deal only become comparable once the GB total is expressed in bits.
Should I evaluate a peering session in gigabytes or in gigabits?
Peering is priced by port and cross-connect, so the working currency is bits per second — what matters is how much of your peak moves off paid transit. Volume still answers the follow-up question. Once you know the GB per month attributable to a peer's prefixes, converting it to gigabits and dividing by the period shows whether that traffic is a steady flow worth a permanent port, or a bursty one better left on transit where the percentile forgives the spikes.
Why do my interface counters and my bandwidth graphs disagree by a factor of eight?
Because SNMP and kernel counters such as ifHCOutOctets accumulate octets — bytes — while the graphing layer conventionally renders link utilisation in bits per second to match the port's rated speed. A dashboard wired straight to the raw counter without the ×8 will draw every value at one-eighth of the truth. It is worth checking the units label on any home-grown egress dashboard before its numbers go into a commit negotiation.
No comments yet. Be the first to comment!